The carbon intensity of grid-supplied electricity depends on the mix of generation sources used to satisfy its demand and varies widely over time and across locations. There are two types of carbon intensity signals: average and marginal. Both signals provide distinct information about grid operations and affect the electric grid’s short- and long-term functioning in different ways. Unfortunately, there is a lack of consensus on the “right” signal for carbon-aware optimizations, and decarbonization efforts across domains have used both signals to decide when and where to shift demand. To understand the implications of signal selection on carbon-aware optimizations, this paper performs a data-driven analysis using both the average and marginal carbon intensity. Our analysis for 65 regions reveals multiple insights, including i) both signals are statistically different with very low correlation between them, ii) optimizing for one signal could lead to more carbon emissions from the other signal’s standpoint, and iii) differences in signal characteristics in each region lead to different electricity use incentives.
24 February 2026 in IEEE Transactions on Computers
As society explores new computing applications, data centers have seen a sharp rise in capacity and energy use, driving up data centers’ carbon footprint and raising concerns about their sustainability. In response, efforts to reduce emissions now complement the traditional goals of cutting energy costs and boosting performance. Given the spatiotemporal variability of the grid carbon intensity, researchers are increasingly adopting workload shifting strategies to minimize the operational carbon footprint of cloud workloads. However, shifting from cost- and performance-centric operations to carbon-aware operations introduces performance penalties and cost overheads. Moreover, large-scale spatial and temporal shifts can cause stampede effects, with synchronized migration to the same low-carbon regions or times, thereby straining resources and creating imbalances. In this paper, we quantify the carbon-cost-performance trade-offs of carbon-aware scheduling and evaluate the risk of such stampede effects. We provide real-world examples illustrating these tradeoffs for application and cloud providers. Lastly, we offer practical insights to help mitigate these challenges and guide sustainable scheduling decisions.